YC 4:2:2 or 4:1:1?

Andre De Clercq · September 25th, 2004, 06:18 AM

Thomas, FYI, DV on itself doesn't generate aliasing which is an effect which occurs when (analog) continious information is being discretized (sampled). So, only the DCT preceding AD conversion process if not well filtered, and the pixel structures in CCD's if there is no optical lowpass filtering, can result in spatio-temporal aliasing effects in the end result. The latter effects are also seen in pure analog camera signals. DV only causes artifacts like musquito noise, quilting/blocking and some complicated motion induced artifacts.

Thomas Smet · September 26th, 2004, 09:40 AM

Hey Jeff I know how the formats and connections work and how they are different. I also know YC isn't as good as component would be. With that in mind however and the "C" being not very good would is still give more color resolution then DV color? Or at least just a good fake of it. I know YC isn't digital so it is hard to compare it with digital. The whole point to this thread was to find out if

1. Any cameras actually gave us raw YC output.
2. If there would be any enhancement to the color information with doing this.

So far the popular opinion has been no even though not very many people have ever done it with a raw YC signal.

Later on in this thread I sort of ditched the YC thing because I cannot test every camera and do not have access to an older analog camera with raw output. I am now trying to figure out if component would be better than DV in terms of color information. Now remember I am not talking about recording to DV tape and then trying to output through component. I am talking about outputting from component live during shooting from the camera.

Some people were saying hands down analog video is not as good as digital and all I was doing was pointing out some areas where that may not be true. Analog component may have a slightly lower resolution but it makes up for it compared to DV with

1. No compression. DV having 5:1 compression.
2. Potential higher color resolution?
3. 10 or 12 bit video. If anybody feels this is worth it or not is an interesting point.

I know there are short comings with analog video but I am trying to find ways around that.

Graeme Nattress · September 26th, 2004, 01:43 PM

Bandwidth of analogue video equates to resolution with digital video. Unless you can measure the bandwidth of the S-Video's chroma, you'll never know for certain. I think, without access to tech notes from the company that makes the camera in question, or to speak to one of their designers, you'll never know. It is most highly likely that the digital to analogue converter in the camera is 8bit, and hence no chance of getting 10 or 12 bit from the S-video.

Graeme

Thomas Smet · September 26th, 2004, 07:26 PM

you might be right Graeme. Since most dv cameras were never meant for this sort of thing I now doubt any of this really matters. I however would still like to know if any advantage would be gained from a raw component output from a camera head. By the time I bought a decent enough camera head with component I might as well get a dvcpro 50 camera.

I have written an e-mail to Black Magic Design the company that makes the 10 bit Decklink capture cards. I asked what their opinion was on the subject.

On the subject of psuedo SD from HDV I did some tests on 1080i HDV footage. No I do not have the SONY HDV camera but I took some Animation from 3D Studio Max and rendered it at an anamorphic 1440 x 1080. I then converted the video to 4:2:2. I have not yet converted to 4:2:0 but will test that later since I need to find a mpeg2 encoder that can do HDV.

I took my 1440x1080i footage at 4:2:2 and scaled down to 720 x 540 and it sort of worked. Instead of smoothing everything out like I thought it would do in theory it kept a lot of jagged edges. It helped a lot but wasn't perfect. I think if we used Graeme's tool first and then scaled down this would help a lot.

Oh by the way nice work on the filter Graeme. I finally had a chance to take a look at it. Do you plan on keeping support for only Final Cut Pro? I am on the PC side and I have been working on a chroma upsampler but I gave it up for now to work on other things.

Bob Cetti · September 26th, 2004, 08:00 PM

Interesting discussion.

Thomas you could check out www.pixelmonger.com. There is an article by Scott Billups from DV magazine. "Re-creating Krakatoa" where he talks about shooting with an analog Beta-SP camera with coponent outputs using direct to disk recording, in order to get D-Beta quality on the cheap for the visual fxs he needed for the project. This could answer your question.

Graeme regarding the high-precision rendering in Final Cut Pro. I have an idea for a production workflow that preserves the DV-25 color resolution as much as possible. This workflow would involve shooting in DV PAL 16:9 with a Canon XL2 in 25p progressive mode. Then firewire transfer to the computer. Do the editing, fxs, color correction, filters, etc. but no rendering. Then when ready to output move the sequence data to an uncompressed sequence format and render in high precision. Then if there a transfer to film just send the hard drives or use this uncompressed file for output to videotape or encoding to dvd, etc. As far as mini-DV goes would this preserve the luma and color resolution as much as possible versus going back to the mini-DV format on the master?

Graeme Nattress · September 27th, 2004, 04:23 AM

Bob, if you edit as DV as normal, then at the end change the sequence settings to that of an uncompressed and render, you will stop any losses incurred by rendering to DV codec. I'm not currently keen on the high precision YUV rendering or 10bit rendering in FCP as it seems a bit buggy to me - you'll have to watch each effect to make sure it renders right.

Graeme

Bob Cetti · September 27th, 2004, 11:14 AM

Thanks for the info Graeme.

As far as green screen, fxs work and FCP filters with DV do you think it would be an advantage to shoot your footage and work with the 4-2-0 color space of PAL versus the 4-1-1 colorspace of NTSC DV?

Graeme Nattress · September 27th, 2004, 11:40 AM

They're both as bad as each other. Just make sure that your chromakey software works on DV. DVMatte is good for that!

Graeme

Barry Green · September 27th, 2004, 03:48 PM

Yeah, they're the same but different.

Adam Wilt did an article on 4:2:0 vs. 4:1:1 in DV magazine, you can probably find it by searching the past articles.

September 26th, 2004, 01:43 PM	#33
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	Bandwidth of analogue video equates to resolution with digital video. Unless you can measure the bandwidth of the S-Video's chroma, you'll never know for certain. I think, without access to tech notes from the company that makes the camera in question, or to speak to one of their designers, you'll never know. It is most highly likely that the digital to analogue converter in the camera is 8bit, and hence no chance of getting 10 or 12 bit from the S-video. Graeme __________________ www.nattress.com - filters for FCP

September 26th, 2004, 08:00 PM	#35
Bob Cetti New Boot Join Date: Oct 2001 Location: Florida Posts: 13	Interesting discussion. Thomas you could check out www.pixelmonger.com. There is an article by Scott Billups from DV magazine. "Re-creating Krakatoa" where he talks about shooting with an analog Beta-SP camera with coponent outputs using direct to disk recording, in order to get D-Beta quality on the cheap for the visual fxs he needed for the project. This could answer your question. Graeme regarding the high-precision rendering in Final Cut Pro. I have an idea for a production workflow that preserves the DV-25 color resolution as much as possible. This workflow would involve shooting in DV PAL 16:9 with a Canon XL2 in 25p progressive mode. Then firewire transfer to the computer. Do the editing, fxs, color correction, filters, etc. but no rendering. Then when ready to output move the sequence data to an uncompressed sequence format and render in high precision. Then if there a transfer to film just send the hard drives or use this uncompressed file for output to videotape or encoding to dvd, etc. As far as mini-DV goes would this preserve the luma and color resolution as much as possible versus going back to the mini-DV format on the master? __________________ Bob Cetti Audio Video Services

September 27th, 2004, 04:23 AM	#36
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	Bob, if you edit as DV as normal, then at the end change the sequence settings to that of an uncompressed and render, you will stop any losses incurred by rendering to DV codec. I'm not currently keen on the high precision YUV rendering or 10bit rendering in FCP as it seems a bit buggy to me - you'll have to watch each effect to make sure it renders right. Graeme __________________ www.nattress.com - filters for FCP

September 27th, 2004, 11:14 AM	#37
Bob Cetti New Boot Join Date: Oct 2001 Location: Florida Posts: 13	Thanks for the info Graeme. As far as green screen, fxs work and FCP filters with DV do you think it would be an advantage to shoot your footage and work with the 4-2-0 color space of PAL versus the 4-1-1 colorspace of NTSC DV? __________________ Bob Cetti Audio Video Services

September 27th, 2004, 11:40 AM	#38
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	They're both as bad as each other. Just make sure that your chromakey software works on DV. DVMatte is good for that! Graeme __________________ www.nattress.com - filters for FCP

September 25th, 2004, 06:18 AM	#31
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Thomas, FYI, DV on itself doesn't generate aliasing which is an effect which occurs when (analog) continious information is being discretized (sampled). So, only the DCT preceding AD conversion process if not well filtered, and the pixel structures in CCD's if there is no optical lowpass filtering, can result in spatio-temporal aliasing effects in the end result. The latter effects are also seen in pure analog camera signals. DV only causes artifacts like musquito noise, quilting/blocking and some complicated motion induced artifacts.

September 26th, 2004, 09:40 AM	#32
Thomas Smet Trustee Join Date: Mar 2004 Location: Milwaukee, WI Posts: 1,719	Hey Jeff I know how the formats and connections work and how they are different. I also know YC isn't as good as component would be. With that in mind however and the "C" being not very good would is still give more color resolution then DV color? Or at least just a good fake of it. I know YC isn't digital so it is hard to compare it with digital. The whole point to this thread was to find out if 1. Any cameras actually gave us raw YC output. 2. If there would be any enhancement to the color information with doing this. So far the popular opinion has been no even though not very many people have ever done it with a raw YC signal. Later on in this thread I sort of ditched the YC thing because I cannot test every camera and do not have access to an older analog camera with raw output. I am now trying to figure out if component would be better than DV in terms of color information. Now remember I am not talking about recording to DV tape and then trying to output through component. I am talking about outputting from component live during shooting from the camera. Some people were saying hands down analog video is not as good as digital and all I was doing was pointing out some areas where that may not be true. Analog component may have a slightly lower resolution but it makes up for it compared to DV with 1. No compression. DV having 5:1 compression. 2. Potential higher color resolution? 3. 10 or 12 bit video. If anybody feels this is worth it or not is an interesting point. I know there are short comings with analog video but I am trying to find ways around that.

September 26th, 2004, 07:26 PM	#34
Thomas Smet Trustee Join Date: Mar 2004 Location: Milwaukee, WI Posts: 1,719	you might be right Graeme. Since most dv cameras were never meant for this sort of thing I now doubt any of this really matters. I however would still like to know if any advantage would be gained from a raw component output from a camera head. By the time I bought a decent enough camera head with component I might as well get a dvcpro 50 camera. I have written an e-mail to Black Magic Design the company that makes the 10 bit Decklink capture cards. I asked what their opinion was on the subject. On the subject of psuedo SD from HDV I did some tests on 1080i HDV footage. No I do not have the SONY HDV camera but I took some Animation from 3D Studio Max and rendered it at an anamorphic 1440 x 1080. I then converted the video to 4:2:2. I have not yet converted to 4:2:0 but will test that later since I need to find a mpeg2 encoder that can do HDV. I took my 1440x1080i footage at 4:2:2 and scaled down to 720 x 540 and it sort of worked. Instead of smoothing everything out like I thought it would do in theory it kept a lot of jagged edges. It helped a lot but wasn't perfect. I think if we used Graeme's tool first and then scaled down this would help a lot. Oh by the way nice work on the filter Graeme. I finally had a chance to take a look at it. Do you plan on keeping support for only Final Cut Pro? I am on the PC side and I have been working on a chroma upsampler but I gave it up for now to work on other things.

September 27th, 2004, 03:48 PM	#39
Barry Green Barry Wan Kenobi Join Date: Jul 2003 Location: North Carolina Posts: 3,863	Yeah, they're the same but different. Adam Wilt did an article on 4:2:0 vs. 4:1:1 in DV magazine, you can probably find it by searching the past articles.